# How AI Helps Solve Medical Mysteries at Boston Children’s Hospital | OpenAI Forum

## Executive summary

Researchers at Boston Children’s Hospital's Manton Center for Orphan Disease Research demonstrated how AI-driven workflows can significantly accelerate the diagnosis of rare genetic diseases. Using models like OpenAI o3 Deep Research, the team analyzed complex genomic data (e.g., whole genome sequencing) and clinical phenotypes to surface novel leads. In one study involving 376 cases, this process led to 18 diagnoses of rare diseases, highlighting AI's potential to transform the 'diagnostic odyssey' by efficiently narrowing massive datasets for expert review.

## Key takeaways

- AI accelerates diagnosis from vast data sets: The workflow uses LLMs to intersect genetic variants (e.g., denovo mutations, deletions/duplications) with curated clinical metadata (ontological codes for phenotypes). This dramatically reduces the search space of thousands of potential variants down to a small, focused list for human diagnosticians.
- AI successfully identified rare diagnoses: In a study across 376 cases, the AI-driven workflow surfaced evidence that led to 18 confirmed diagnoses of rare diseases. The model can also suggest gene-phenotype associations based on literature even when those links were previously unknown.
- The diagnostic process is iterative and requires human oversight: While the AI accelerates analysis, it does not replace the diagnostician. The model's output must be reviewed by experts who validate the evidence-driven list of hypotheses to ensure accuracy and guide follow-on testing.

## Technical details

- Genomic Data Analysis: The process involves analyzing whole genome sequencing data (e.g., a dataset containing potentially 500,000 rows and 20-30 columns of information per genetic change). Analysts filter out common population variants to focus on rare or novel changes.
- LLM Workflow Integration: The LLM is prompted to intersect knowledge by correlating clinical presentations (defined by numerical ontological codes) with genetic changes. This process aims to nominate a small number of potential genes or variants for expert review.
- Model Development and Validation: The model was initially tested on known, solved cases (e.g., giving it 20 solved cases to get 19 correct) to refine prompting strategies and prevent common technical errors (like mistaking a false positive for a true variant).

## Practical implications

- Accelerating the diagnosis of rare diseases by providing a systematic method to analyze complex genetic and clinical data.
- Enabling personalized medicine approaches by identifying underlying genetic causes rather than just treating symptoms.
- Creating an accessible tool that can be periodically rerun on existing cases, allowing for new diagnoses as scientific knowledge evolves.

## Topics

Artificial Intelligence (AI), Genomics, Rare Diseases, Computational Biology, Healthcare Informatics, OpenAI Forum, Manton Center for Orphan Disease Research

Source: https://www.youtube.com/watch?v=SJJBPwlPpww